Mean-Shift of Variable Window Based on the Epanechnikov Kernel

摘要

Mean shift is an effectively iterative algorithm, which has been widely used in image processing and computer vision. The monotonicity of the density based on the mean shift of the variable window and the convergence of the mean shift produce are not been strictly proved. The convergence of mean shift produce is proved according to Cauchy Convergence Criterion. The whole increasing of the density according to the Epanechnikov kernel is proved.